

The file data.xls contains the US cities CPI data, data.xls, and the GAUSS .prg and .src
code files to compute the KPZ W_1a and W_1b tests, written by Mohitosh Kejriwal, publicly
available at Pierre Perron's codes depository web page http://blogs.bu.edu/perron/codes/

Please, note that the GAUSS codes are not of my own.

To run the .prg's as provided in this file, just store the difference of a city CPI in logs, with
respect the log of Chicago CPI in a txt file "y.txt". You can however input the data in different ways just changing the
corresponding line in the .prg code files.

The rest of procedures of the paper have been computed using EViews, version 9, as follows.

To compute the Bai-Perron test:

1.Load the Excel data file
2.Take logs of each series and difference with respect the log of Chicago
3.For a given individual difference series (e.g., log(Baltimore) with respect of log(Chicago)), run the ADF regression
with maximum lag selected by Modified Akaike info criterion (MADF) 
4.Run the BP Multiple Breakpoint Test of the Stability Diagnostics menu	(Method: Global breaks vs. none), allowing error distributions
to differ across breaks and deleting the lagged first differences in the Regressor to vary across breakpoints list

To compute the MZ test:

Given the breakpoint(s) of the KPZ test for a given individual difference series, run an ADF regression
with maximum lag chosen by MAIC for each sub sample. For the sub sample with the largest estimated autorregresive root,
compute the Ng-Perron test with lag parameter chosen as in Perron and Qu.

If additional help is needed, please do not hesitate to contact me at jorge.belaire@uv.es


